K-means online-learning routing protocol (K-MORP) for unmanned aerial vehicles (UAV) adhoc networks

Abstract

Unmanned Aerial Vehicles (UAVs) have become a hot topic due to their flexible architecture adopted in many wireless technologies. In UAV ad hoc networks, traditional routing protocols with a fixed topology are ineffective due to dynamic mobility and unstable paths. Therefore, the mobility patterns of UAVs challenge efficient and reliable routing in UAV networks. Traditional routing algorithms are often based on assumptions of static nodes and predetermined network topologies. Which are not suitable for the dynamic and unpredictable nature of UAV mobility patterns. To address this problem, this paper introduces a K-means online learning routing protocol (KMORP) scheme employing a Markov mobility model for UAV ad hoc networks. Initially, the proposed method utilizes a 3D Gauss Markov mobility model to accurately estimate UAV positions, while K-means online learning is adopted for dynamic clustering and load balancing. Designed for real-time data processing, KMORP is well suited for UAV ad hoc networks, quickly adapting to network environmental changes such as UAV mobility, interference, and signal degradation to ensure efficient data transmission and communication. This is achieved while reducing the overall communication overhead and increasing the packet delivery ratio(PDR%). In the routing phase, the proposed scheme employs inter-cluster forwarding nodes to transmit messages among different clusters. Extensive simulations demonstrate the performance of the proposed KMORP, showing a 38% better PDR compared to OLSR and over 50% less end-to-end(E2E) delay compared to typical K-Means. Furthermore, the proposed KMORP exhibited an average throughput of 955 kbps, showing a substantial improvement in network performance. The results underscore that the proposed KMORP outperforms existing techniques in terms of PDR, E2E delay, and throughput.